The ability to rationally design small molecules that bind with high af?nity and speci?city to one or more biomolecu- lar targets would radically transform drug discovery. Current approaches require many rounds of screening, mod- eling, and synthesis in a trial-and-error approach that is costly, time-consuming, and ineffective. After decades of work on the study of biomolecular interactions, there remains an enormous gulf between what we claim to un- derstand about biomolecular association and our ability to put this knowledge into practice. This gulf is especially wide for the design of selective kinase inhibitors, which aim to target one or more speci?c kinases in order to effectively treat a disease?often cancer?and minimize unwanted toxic side effects. While the discovery of imatinib was hailed as a breakthrough for its ability to selectively inhibit Abl despite the existence of closely related kinases like Src, it came as a great surprise when the crystal structure of imatinib bound to Src revealed that the Src:imatinib complex was nearly identical to Abl:imatinib. Recent evidence from both experiments and modeling has suggested that a previously underappreciated contribution?the energetic cost of populating the inhibitor-bound conformation?plays a critical role in the selectivity of imatinib for Abl over Src. While this effect has only been studied in the well-studied case of Abl/Src binding to imatinib, it has the potential to be much more general. We hypothesize that exploiting differences in the energetic cost of con?ning related kinases to inhibitor binding-competent conformations may be a route to selectivity in targeted kinase inhibition. Here, we ask how much conformational reorganization energy contributes to the af?nity of current FDA-approved noncovalent kinase inhibitors to determine whether existing inhibitors exploit differences in these reorganization energies (perhaps inadvertently) to achieve selectivity, and whether there is a clear route to exploiting this difference in rationally engineering new selective molecules. We use a combined experimental and computational approach to decompose inhibitor binding af?nity and se- lectivity into contributions from kinase reorganization and binding to individual kinase conformations. We ?rst computationally map the conformations accessible to a diverse panel of human kinase catalytic domains, along with their associated energetics. By using an automated ?uorescence assay to measure the af?nities of FDA- approved noncovalent inhibitors to this panel and alchemical free energy calculations to determine the inhibitor binding af?nities to individual conformations, we can combine these data to quantify the relative contribution of reorganization energy to the af?nity and selectivity of kinase inhibition. We then use the introduction of point mutants intended to modulate selectivity via reorganization energies to validate our model, and examine oppor- tunities for exploiting differences in reorganization energy between related kinases or wild-type and mutationally activated kinases as a route to selectivity.